The mandate memo

Shopify’s CEO wrote it as an internal memo, and on April 7 it went public. It made “reflexive AI usage” a baseline expectation for everyone, said AI use would show up in performance reviews, and asked teams to show why AI couldn’t do a job before asking for headcount.

Most design leaders will get a version of that memo this year, from their own CEO, in their own words. Decide now what you’ll do when it lands.

What a mandate can do

A mandate is good at one thing: ending the debate about whether to try. That counts for a lot. Plenty of teams spent 2024 waiting for permission, or for the tools to get good, or for someone else to go first. A memo from the top settles that in a day.

What a mandate can’t do is tell you whether the work got better. Usage is easy to measure: seats, prompts, minutes. It’s also the wrong number. A team can use AI all day and ship the same work faster, or worse work at the same speed, and the dashboard looks identical.

Measure the work instead

For a design team there’s one question a usage report can’t answer: is what we shipped better than what we’d have shipped last year?

You can check that without a new dashboard. Put this quarter’s shipped work next to last year’s in a critique, the same surfaces where you can, and have someone senior say which is better and why. Do it every quarter. It’s slower than counting prompts, and of everything on the list it’s the measure that tells you whether the tools helped.

Answering usage with usage

The easy response to a usage mandate is more usage: count the prototypes, post the demos, report the adoption rate. It looks like leadership and changes nothing.

The better answer is to raise the bar in the same breath. If the tools make making cheap, ask for more: more directions explored before committing, more rigor on behavior, more care on the details that generated work gets wrong, because generated work comes back as the average unless someone holds it to more.